I am trying to solve the Laplace Equation on the rectangle. Namely, I trying to solve the following PDE problem:

Solve the equation $\Delta u = 0$ on the rectangle $R = \{(x, y): 0\leq x \leq a, \hspace{2mm} 0\leq y \leq b\}$ subject to $u(x,0) = f(x)$, $u(x, b) = 0$, $u(0, y) = 0$ and $u(a, y) = 0$.

I have separated variables and used the first three boundary conditions to obtain

$$u(x, y) = \sum_{n= 1}^{\infty} c_n \sin\left(\frac{n\pi x}{a}\right)\sinh\left(\frac{n\pi (b-y)}{a}\right)$$

Now it is the final boundary condition that seems strange. I am reading how it has been dealt with Olver's introduction book to PDEs. His argument goes as follows:

We want

$$u(x, 0) =:f(x) = \sum_{n= 1}^{\infty} c_n \sinh\left(\frac{n\pi b}{a}\right)\sin\left(\frac{n\pi x}{a}\right)$$

and this looks like the fourier sine series for $f(x)$.

So let $$b_n = \frac{2}{a} \int_0^a f(x) \sin\left(\frac{n\pi x}{a}\right)dx$$ then $c_n = \frac{b_n}{\sinh\left(\frac{n\pi b}{a}\right)}$.

But the issue I have here is that does the fact that $f$ has a fourier sine series not assume/imply that the function $f$ is odd? How is this possible - was the choice of $f$ not arbitrary? So the PDE problem could have been posed with a specific function for $f$ where $f$ is instead even and then this does not make sense anymore?


The point is that you only care about the values that $f$ takes for $0 \leq x \leq a$: you can choose the other values however you like, or not at all. All the book is doing is representing $f$ as a sine series on the interval $[0,a]$. It so happens that said sine series extends naturally to an odd function with period $2a$, but that this happens is completely irrelevant for the properties of the solution of the equation in the region considered.

  • $\begingroup$ thanks, really helpful and clear $\endgroup$ – PhysicsMathsLove May 28 '18 at 15:57

You can extend $f$ to an odd function on $[-a,a]$ and expand in a sine series. If $f$ has left- and right- derivatives at $0$, then the resulting sin series will converge to $0$ at $x=0$ and at $x=a$, which may not match up with $f$ at those points. This is not unreasonable because consistency in the conditions for the PDE would require that $f(0)=f(a)=0$ because $u(0,y)=0$, $u(a,y)=0$ are also required to hold for $y$. In any event, the Fourier sine series will converge in $L^2[0,a]$ to the prescribed function.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.